[csw-maintainers] hudson (previously, git in testing)

[Trygve Laugstøl]

Philip Brown wrote:
> On Thu, Dec 18, 2008 at 07:54:02PM -0500, Ben Walton wrote:
>> Excerpts from Philip Brown's message of Thu Dec 18 18:42:30 -0500 2008:
>>> but in what environment? unless it's a bottom-level library, it has to get
>>> what it "makes" against, from SOMEWHERE.  does hudson, not get that stuff
>>> from other things "in hudson" ??
>> You're right Phil, in that adding git to hudson would currently be
>> useless since it can't use the dependencies that don't exist yet.  For
>> packages that do have all dependencies published, I don't see any harm
>> in having hudson ensure they're continuing to build (think gar
>> changes, etc)...Even if the package doesn't get better (more
>> platforms, for example), hudson can make sure it doesn't get worse.
>>
> 
> dependson what the purpose of hudson is.
> If it's just "play around and see if things build nicely", then great.
> 
> 
> If on the other hand, it is on the road to becoming
>  "the official way to build packages before publication", 
> as some people seem to wish, then there needs to be stricter rules about
> putting things into hudson, otherwise we risk massive corruption.

That is right, but right it is working in the first way. I would like to 
extend that way to automatically pkgrm/pkgadd the packages once they're 
done.

I'd like to start the second way once we have some more experience with 
Hudson and GAR has the few glitches ironed out (and we have a lot of 
time to actually get the process going).

--
Trygve